Inclusion of 30-day postdischarge detection triples the incidence of hospital-onset methicillin-resistant Staphylococcus aureus

BACKGROUND: Hospitalized patients are at increased risk for acquisition of methicillin-resistant Staphylococcus aureus (MRSA). As hospital length of stay shortens, hospital-acquired MRSA events may be more likely to be detected after discharge.

OBJECTIVE: We assessed the impact of attributing MRSA cases discovered within 30 days after discharge to the most recent hospitalization and identified patient characteristics associated with MRSA detection after discharge.

DESIGN: Retrospective cohort study.

SETTING: Twenty-seven acute care hospitals in Orange County, California.

PARTICIPANTS: Adult acute care admissions (2002-2007).

METHODS: Using a countywide hospital data set containing diagnostic codes with present-on-admission (POA) indicators, we identified the first admission with a MRSA code for each patient. This incident MRSA admission was defined as predischarge-detected (pre-DD) hospital-onset MRSA (HO-MRSA) when MRSA was not POA. If MRSA was POA and a prior admission occurred within 30 days, this prior admission was assigned postdischarge-detected (post-DD) HO-MRSA. We evaluated the impact of including post-DD HO-MRSA in the calculation of hospital HO-MRSA incidence using signed-rank tests and reviewed changes in hospital rankings. We conducted multivariate comparisons of patient characteristics of pre-DD versus post-DD HO-MRSA patients.

RESULTS: Among 1,217,253 at-risk hospitalizations, the inclusion of post-DD HO-MRSA tripled the median hospital HO-MRSA incidence, from 12.2 to 35.7 cases per 10,000 at-risk admissions (P < .0001). Hospital ranking changed substantially when including post-DD HO-MRSA. Patients with shorter stays were more likely to have post-DD MRSA.

CONCLUSIONS: On the basis of administrative claims data, the inclusion of post-DD HO-MRSA significantly increased the estimated HO-MRSA incidence and altered hospital rankings. This finding underscores the limitations of single-facility data when deriving HO-MRSA incidence and rank.